Machine Learning-Based Anomaly Detection for ITER's Tokamak Systems Monitor: A Gyrotron Case Study

Abstract

The ITER Tokamak Systems Monitor will provide operators with timely assessments of machine health, component lifetime, and early warnings of potential faults. This proof of concept applies dimensionality reduction and clustering to gyrotron pulses, identifying deviations from expected operational patterns. The approach is designed for intershot anomaly detection and can be adapted to other systems and signals within the Tokamak Systems Monitor.

Publication
In IEEE Transactions on Plasma Science